Matizmo

Your campaign just went live. The social creative looks sharp. The email header is clean. The ad variants are consistent. And none of it is legally yours.

That is not a hypothetical. It is the current state of copyright law in the United States, and most cybersecurity marketing teams have no idea it applies to them.

What the courts have actually decided

In March 2025, the U.S. Court of Appeals for the District of Columbia issued its ruling in Thaler v. Perlmutter, affirming that the Copyright Act of 1976 does not protect works created entirely by artificial intelligence. Writing for the unanimous panel, Judge Patricia Millett was direct: "Authors are at the centre of the Copyright Act." Machines cannot own property, have a lifespan, or provide an authenticating signature. Therefore, a machine cannot be an author. The Supreme Court declined to hear an appeal in March 2026, leaving the ruling intact.

The court was careful to note that it was not ruling on AI-assisted work, only on work where AI was listed as the sole author. But that nuance is cold comfort for most marketing teams. The Copyright Office's January 2025 guidance makes clear that protection depends on whether "a human author has determined sufficient expressive elements" in the output. Prompts alone may not be enough. In a parallel case, Allen v. Perlmutter, the Copyright Office denied registration to an artist who had used more than 600 carefully crafted Midjourney prompts to create an award-winning image, ruling that he had "no control over how the artificial intelligence tool analysed, interpreted, or responded to these prompts."

"No version logs, no documented human creative contribution, no IP ownership over your ads, social creative, or email designs."

The practical implication most teams are missing

Here is what this means in practice. If your team generates campaign assets using an AI tool, publishes them without maintaining version logs or documented evidence of human creative contribution, and a competitor copies those assets, you may have no legal recourse. The work is not copyrightable. It is, in effect, in the public domain.

This is not a theoretical edge case. It is the default outcome for any team that treats AI image generation as a production shortcut rather than a documented creative process.

The risk gets worse when you consider the tool you are using. Not all AI platforms carry the same legal exposure.

Adobe Firefly versus Midjourney: a meaningful difference

Adobe offers contractual IP indemnification to enterprise customers on qualifying plans. According to Adobe's Firefly Legal FAQs for Enterprise Customers, the indemnity covers claims that allege Firefly output "directly infringes or violates any third party's patent, copyright, trademark, publicity rights or privacy rights." Adobe trained Firefly on licensed content and public domain material specifically to make this indemnity commercially viable.

Midjourney offers no equivalent protection. Its Terms of Service grant paid subscribers the right to use outputs commercially, but Midjourney makes no warranty about the copyright status of those outputs and provides no indemnification if a claim arises. The platform is currently the subject of copyright infringement lawsuits from major studios including Warner Bros., Disney, and Universal over its training data. If those suits succeed, the downstream implications for commercial users of Midjourney outputs are unclear.

Tool	IP Indemnification	Training data	User liability if claim arises
Adobe Firefly (enterprise)	Yes, contractual	Licensed and public domain	Covered under Adobe indemnity
Midjourney (any plan)	No	Disputed	User bears full liability
Canva AI (enterprise)	Partial, via Adobe Firefly integration	Mixed	Varies by feature

This is not an argument against using Midjourney. It is an argument for knowing what you are using and building your process accordingly.

What a defensible AI creative process looks like

The Copyright Office's position points to a clear standard: documented human creative contribution. That means keeping records of the decisions a human made in the creative process, not just the final output. In practice, this means:

Maintaining version logs that show prompt iterations and human selection decisions
Keeping records of any post-generation editing, retouching, or layout work applied to AI outputs
Ensuring that a human creative director or designer can articulate the expressive choices they made in directing and selecting the final asset

None of this is onerous. It is the kind of documentation that good creative teams already produce. The difference is that for AI-generated work, it is no longer optional if you want to own what you make.

"The question is not whether you are using AI. The question is whether you can prove a human made the creative decisions."

The question worth asking before your next campaign

Most cybersecurity marketing teams think about output quality and production speed when they evaluate AI tools. Legal exposure rarely comes up until something goes wrong.

It is worth asking now: does your team keep version logs for AI-generated campaign assets? Do you have a documented process for establishing human authorship over AI-assisted creative work? Do you know which tools in your stack carry IP indemnification and which do not?

If the answer to any of those is no, the assets you are producing today may be freely reusable by anyone, including your competitors, the moment they go live.

We can help you build an AI design system with the governance to match.

References: